app.py

import os
import gradio as gr
import requests
import pandas as pd
from agent import GaiaAgent

DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"

def run_and_submit_all(profile: gr.OAuthProfile | None):
    """
    Fetch questions from GAIA API, run agent, submit answers
    """
    
    # Check login
    if not profile:
        return "Please login to Hugging Face first.", None
    
    username = profile.username
    space_id = os.getenv("SPACE_ID")
    
    print(f"\n[run_and_submit_all] starting for user: {username}")
    
    # Fetch questions from API
    print("[run_and_submit_all] fetching questions from API...")
    try:
        questions_resp = requests.get(f"{DEFAULT_API_URL}/questions", timeout=30)
        questions_resp.raise_for_status()
        questions = questions_resp.json()
        print(f"[run_and_submit_all] ✓ fetched {len(questions)} questions")
    except Exception as e:
        error_msg = f"Error fetching questions: {str(e)[:200]}"
        print(f"[run_and_submit_all] ✗ {error_msg}")
        return error_msg, None
    
    # Initialize agent
    print("[run_and_submit_all] initializing agent...")
    agent = GaiaAgent()
    
    # Run agent on each question
    results_log = []
    answers_payload = []
    
    for i, item in enumerate(questions):
        task_id = item.get("task_id")
        question = item.get("question")
        file_name = item.get("file_name", "")
        
        print(f"\n[run_and_submit_all] [{i+1}/{len(questions)}] task_id={task_id}")
        print(f"  question: {question[:80]}...")
        print(f"  file: {file_name if file_name else '(none)'}")
        
        file_content = ""
        
        # Try to fetch file if it exists
        if file_name:
            try:
                print(f"  fetching file: {file_name}...")
                file_resp = requests.get(
                    f"{DEFAULT_API_URL}/files/{task_id}",
                    timeout=30
                )
                if file_resp.status_code == 200:
                    # Try to decode as text first
                    try:
                        file_content = file_resp.text[:5000]
                        print(f"  loaded {len(file_content)} chars from file")
                    except:
                        # If binary, note it
                        file_content = f"[Binary file: {file_name}, {len(file_resp.content)} bytes]"
                        print(f"  loaded binary file")
                else:
                    print(f"  file fetch returned {file_resp.status_code} (skipping)")
            except Exception as e:
                print(f"  error fetching file: {e}")
        
        # Run agent
        try:
            answer = agent(question, file_content=file_content)
        except Exception as e:
            print(f"  error running agent: {e}")
            answer = "I am unable to answer"
        
        answers_payload.append({"task_id": task_id, "submitted_answer": answer})
        results_log.append({
            "Task ID": task_id[:8] + "...",
            "Question": question[:60] + "...",
            "Answer": answer[:60] + "..." if len(answer) > 60 else answer
        })
    
    # Submit answers
    print(f"\n[run_and_submit_all] submitting {len(answers_payload)} answers...")
    submission_data = {
        "username": username.strip(),
        "agent_code": f"https://huggingface.co/spaces/{space_id}/tree/main" if space_id else "local",
        "answers": answers_payload
    }
    
    try:
        response = requests.post(
            f"{DEFAULT_API_URL}/submit",
            json=submission_data,
            timeout=60
        )
        response.raise_for_status()
        result = response.json()
        
        status_msg = (
            f"✅ Submission Successful!\n"
            f"User: {result.get('username')}\n"
            f"Score: {result.get('score', 'N/A')}% "
            f"({result.get('correct_count', '?')}/{result.get('total_attempted', '?')} correct)\n"
            f"Message: {result.get('message', 'No message')}"
        )
        print(status_msg)
        return status_msg, pd.DataFrame(results_log)
        
    except Exception as e:
        error_msg = f"❌ Submission failed: {str(e)[:200]}"
        print(error_msg)
        return error_msg, pd.DataFrame(results_log)


# Gradio UI
with gr.Blocks() as demo:
    gr.Markdown("# GAIA Agent — Mistral")
    gr.Markdown("""
    **How it works:**
    1. Click "Login with Hugging Face"
    2. Click "Run Evaluation"
    3. Agent processes all 20 questions
    4. See your score instantly!
    
    **Features:**
    - Uses Mistral model via Groq API
    - Web search via DuckDuckGo (free, no keys)
    - Fetches files from GAIA API
    - Automatic answer submission
    """)
    
    gr.LoginButton()
    run_btn = gr.Button("Run Evaluation & Submit", size="lg", variant="primary")
    
    status_output = gr.Textbox(label="Status / Result", lines=5, interactive=False)
    results_table = gr.DataFrame(label="Results", wrap=True)
    
    run_btn.click(
        fn=run_and_submit_all,
        outputs=[status_output, results_table]
    )

if __name__ == "__main__":
    demo.launch(debug=True, share=False)